What is the outlook for Julia?

5 points by tlmr 11 years ago | 11 comments

TLDR: Will the Julia grow to a significant fraction of python's popularity and capability for general, web and data science programming?

I was looking at python for a good general purpose scripting language, .net libraries, etc with statistical and data science capabilities.

It seems that despite the success and burgeoning capabilities, The pydata ecosystem has inherent limitations (that I would bump up against in my use cases).

So I turn to Julia a purportedly general purpose language that excels at numerics, with easy python interopt to ameliorate the current embryonic state of libraries.

Sounds great, but the analyst in me is looking for one language that is versatile while still fullfililng my data analysis requirements... I don't want to invest in an ecosystem that will stagnate in terms of general programming, web programming, data science and job demand.

I feel bullish on Julia, but not completely sure. I would be quite grateful for everyone's thoughts on this?

Thanks

Fomite 11 years ago |

The CrossValidated site on Stack Exchange, which is devoted to statistics and machine learning, had a question on this with some solid answers: http://stats.stackexchange.com/q/25672/5836

My take is that I'd really like Julia to take off (I would really like to be less penalized for writing loops), but libraries, and especially libraries that allow people to use it as an analysis language (rather than for writing their own bespoke stuff) is essential.

tlmr 11 years ago | |

Thanks. What do you think of pycall?

Fomite 11 years ago | | |

Honestly, I haven't used it. Julia is a curiosity for me at the moment, and my workload (I'm an epidemiologist, and there's a bit of an Ebola crisis at the moment) doesn't allow for experimenting with new languages at the moment enough to have delved deeply into interoperability between Python and Julia.

sharmi 11 years ago |

I have been working with python for years. I tried Julia a few months back. The core language is well designed but the surrounding libraries are still far from mature. I found myself dipping into python through pycall for basic needs like URL downloading. That said, I will be going back again :-) The community is very responsive. It used to make me wonder whether the core team ever took a break. TLDR if you are willing to contribute where Julia lacks, pl go ahead. Else, if you are looking for a stable environment, you are better off with python.

tlmr 11 years ago | |

Appreciate your input. Sounds like I am better off sticking with python for the time being.

kylebgorman 11 years ago |

If Julia catches on, it will mostly cannibalize R, not Python.

antman 11 years ago |

I dont think they can yet interopt. I would try zeromq there is some example code for both. Julia is improving rapidly and it has lots of tools, but only for data analysis. If you want create a web service for exampe you are better off with Python.

ihnorton 11 years ago | |

PyCall has been used for a significant range of Python packages (from libraries like scikit and NLTK to PyQt).

https://github.com/stevengj/PyCall.jl

antman 11 years ago | | |

I meant that the problem is calling Julia from Python. Thus the webservice example. Python makes it easier to have a frontend, in flask or django and it would be nice if you could call Julia background services easier.

tlmr 11 years ago | | |

What about beautiful soup?

tlmr 11 years ago | |

By interopt, I meant calling python functions like they are julia functions.